The development of structure and function in the nervous system is the product of stereotyped cell divisions and cell movements, that generate a complex, interconnected system out of an apparently homogeneous neural tube. In many respects, developmental neurobiology of vertebrates remains in its infancy, with many unanswered basic mechanistic questions. The intrinsic difficulty in addressing mechanistic questions is compounded by the paucity of structural data available on the developing vertebrate brain. Without knowing what is happening (as well as where and when), it is very difficult to formulate and evaluate hypotheses about how events come about. The specific aims of this project address the fundamental what, when, and where aspects of brain development: . GOAL DIRECTED IMAGING- Provide comprehensive three-dimensional in vivo images of the developing brain in three different species (qual, mouse, and lemur) using Magnetic Resonance (MR) micro-imaging. Correlate information from functional MRI with structural information from atlas data. . ANALYSIS- Analyze the images at a number of different levels ranging from simple annotation to correlating the onset and developing organization of different structures in both space and time. Compare and contrast the avian, rodent, and mammalian systems. . DISSEMINATION- Make the images and tools for rendering and analysis available to the scientific community in convenient formats ranging from photographs of 2D slices of the 3D data to CD-ROMs with annotated 3D data sets and WWW viewing tools. All current atlases of development take the form of a series of 2 dimensional micrographs of histologically processed and thin sectioned specimens. MR imaging provides 3 dimensional images of thick opaque samples in a non-invasive manner, obviating the need to sacrifice/fix/stain/section a specimen in order to image it. Thus, it is straightforward to perform longitudinal studies of structure and pattern formation by repeated MR imaging of the same animal as it develops. The images are inherently 'stereotactic' in the sense that distances, areas, volumes, angular orientations, etc. are all readily available from the data.